[
https://issues.apache.org/jira/browse/HDDS-16007?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Hsu Han Wen updated HDDS-16007:
-------------------------------
Description:
A datanode may enter the RUNNING state even though it can never register with
SCM, leaving it appearing to be online while it is actually unusable.
One example is when the configured hostname cannot be resolved. The datanode
continues startup, but later fails to register with SCM because the required
registration information cannot be constructed. This also causes
`MiniOzoneCluster.waitForClusterToBeReady()` to eventually time out.
The datanode should fail fast instead of entering the RUNNING state when it is
known that SCM registration cannot succeed.
Related discussion:
https://github.com/apache/ozone/pull/10598#discussion_r3654759640
was:
A datanode may enter the RUNNING state even though it can never register with
SCM, leaving it appearing to be online while it is actually unusable.
One example is when the configured hostname cannot be resolved. The datanode
continues startup, but later fails to register with SCM because the required
registration information cannot be constructed. This also causes
`MiniOzoneCluster.waitForClusterToBeReady()` to eventually time out.
The datanode should fail fast instead of entering the RUNNING state when it is
known that SCM registration cannot succeed.
Related discussion:
https://github.com/apache/ozone/pull/10598
> Datanode should not enter RUNNING state when it cannot register with SCM
> ------------------------------------------------------------------------
>
> Key: HDDS-16007
> URL: https://issues.apache.org/jira/browse/HDDS-16007
> Project: Apache Ozone
> Issue Type: Bug
> Reporter: Hsu Han Wen
> Priority: Major
>
> A datanode may enter the RUNNING state even though it can never register with
> SCM, leaving it appearing to be online while it is actually unusable.
> One example is when the configured hostname cannot be resolved. The datanode
> continues startup, but later fails to register with SCM because the required
> registration information cannot be constructed. This also causes
> `MiniOzoneCluster.waitForClusterToBeReady()` to eventually time out.
> The datanode should fail fast instead of entering the RUNNING state when it is
> known that SCM registration cannot succeed.
> Related discussion:
> https://github.com/apache/ozone/pull/10598#discussion_r3654759640
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
